How Cheese Was Invented

I have been telling a story about this for years, and I want to start here because I found out it is the interesting kind of wrong.

The story I always believed goes like this. Somewhere back in the deep past, herders butchered a sheep, and somebody thought to use the animal’s stomach as a bag. They filled it with milk, slung it over a shoulder, and went out to do a day’s work. By the time they stopped to drink, the milk had turned — curds in whey, sweet and firm and obviously food. The rennet left in the stomach lining had done it. Cheese, by accident, on the move.

It is a wonderful story. It is also the version you will find on a hundred cheese-shop chalkboards and on the back of a hundred packages, usually with an Arabian merchant and a desert crossing bolted on for colour. So I went looking for where it comes from, and what the evidence actually says.

Here is what I found.

The mechanism is real. The story is not evidence.

Start with the part that holds up, because most of it does.

A ruminant’s fourth stomach — the abomasum — produces chymosin, the enzyme we call rennet, and it is very good at its job. A young animal still on milk makes it in quantity, because the whole point is to clot milk in the gut so it can be digested slowly. Put fresh milk into a container made from that stomach, keep it near body temperature, give it a few hours of jostling, and you will get curd. There is nothing hypothetical about it. That is precisely how rennet was harvested and used for most of the history of cheesemaking, and a dried stomach was standard cheesemaking equipment into living memory.

So could it have happened by accident, exactly the way the story says? Absolutely. Nobody serious disputes it, and it remains the most plausible account of how rennet cheese specifically got started.

But notice what just happened. “This could plausibly have occurred” quietly became “this is how it happened,” and then it grew a merchant, a desert, a saddlebag and a sunny afternoon. I cannot find a primary source for any of that. No text, no tablet, no excavation — it is a tale that gets repeated because it is a good tale and because it is mechanically sound, which makes it feel like a finding. It isn’t one. It’s a reconstruction wearing a costume.

And while we’re auditing details: the sheep is decoration. In the Neolithic Near East, where this would have happened, a kid or a calf stomach is at least as likely a container, and calf rennet is what the trade standardised on later. My version had a sheep in it because the version I heard had a sheep in it.

What the dirt actually says — and it isn’t rennet

Here is the part that genuinely changed how I think about my own cheesemaking.

The oldest hard evidence we have for cheesemaking does not point at a stomach bag at all. It points at a sieve.

In the Kuyavia region of Poland, archaeologists have been digging up perforated pottery — fired clay vessels punched through with small holes — for decades. For more than thirty years the typological read on them was “these are cheese-strainers,” because that is exactly what they look like: they look like the colander I drain curd in. But looking like something is not being something, and the identification sat unproven.

In 2013, a team led by Richard Evershed at Bristol, with colleagues in Poland and at Princeton, settled it (Nature 493, published online at the end of 2012). They pulled the fatty acids still trapped in the fabric of the pottery and ran lipid biomarker and stable isotope analysis on them. The perforated vessels carried dairy residues. Even better, the same assemblage told a coherent kitchen story: the cooking pots held ruminant carcass fats — meat — and the bottles held beeswax, while the sieves and some unperforated bowls held milk. Different vessels, different jobs, one household. That is about 7,000 years ago, in the sixth millennium BC.

Now look at what a sieve is for. A sieve separates curd from whey. It does not tell you how the curd was made — but it tells you that somebody had curd and wanted it drained, and drained fresh curd is the whole family of cheeses I started with. Which brings me to the part I did not expect.

My first cheese was the old way.

When I made mozzarella for the first time in the early nineties, I used raw milk from the farm down the road, vinegar, and flake salt. No rennet — I didn’t own any yet. I have always thought of that as the beginner’s shortcut, the version you do before you get serious and order chymosin. But acid coagulation is not a shortcut around the tradition. On the current evidence it may well be the older branch of it. Souring is free and automatic: leave raw milk somewhere warm and its own bacteria will make the acid for you and drop the curd out with no enzyme and no equipment at all. A sieve is the only tool you actually need.

Whether the earliest cheese was acid-set or rennet-set is genuinely unsettled, and I want to be straight that the evidence does not resolve it. A drained curd is a drained curd; the residue in the clay cannot tell us which mechanism dropped it. But the artefact we can actually date is a strainer, not a stomach — and that is worth sitting with, because the story we all repeat has it the other way round.

Why bother making cheese at all?

This is the question I think matters most, and it has the best answer.

Milk is a problem. It is a nearly perfect food that spoils in hours, arrives on a schedule you don’t control, and shows up in quantity exactly when you least need calories — spring and summer, when everything else is growing too. Cheese solves every part of that at once. It concentrates the fat and protein, throws away the water that microbes need, and turns a perishable daily trickle into something you can stack on a shelf, carry over a mountain, trade, tax, and eat in February. Every single thing I do in my aging fridge is a version of that one trick.

And then there is lactose. Most adult humans in the Neolithic were lactase non-persistent — after weaning they stopped making much of the enzyme that splits milk sugar. Cheesemaking strips most of the lactose out with the whey, and what survives into the curd gets eaten by the bacteria as the cheese ripens. Peter Bogucki, one of the authors on that Polish paper, put it plainly at the time: “Making cheese allowed them to reduce the lactose content of milk, and we know that at that time, most of the humans were not tolerant to lactose.”

So: people couldn’t digest milk, therefore they invented cheese, and later they evolved the gene to drink milk straight. Tidy. That is the version I was going to write, and I would have been repeating a story that the best available research has taken apart.

The correction that made this page worth writing

In 2022, Evershed and a large team published something much bigger in Nature (608:336–345) — roughly 7,000 pottery fat residues from more than 550 archaeological sites, mapping milk use across Europe over 9,000 years. It is the largest dataset anyone has assembled on the question, and three of its findings dismantle the tidy version.

First: milk use was widespread from the Neolithic onward. Not marginal, not a workaround — ordinary, though it varied a lot from place to place and century to century.

Second, and this is the surprise: how much milk people used does not explain the spread of the gene. The team tested whether lactase-persistence selection tracked milk exploitation and reported that it “provides no better explanation of LP allele frequency trajectories than uniform selection since the Neolithic.” More milk did not mean more selection pressure. The intuitive engine of the whole story isn’t there.

Third: the gene doesn’t even do much for modern milk drinkers. In a cohort of about 500,000 living Europeans in the UK Biobank, carrying the lactase-persistence genotype was “only weakly associated with milk consumption” and showed no consistent link to better health or fitness. Plenty of people without it drink milk. Plenty with it don’t.

What the authors propose instead is grimmer and, I think, much more convincing. Lactase non-persistent people drank milk anyway — being non-persistent means discomfort, not incapacity, and how much discomfort varies enormously between people. Most of the time, in a normal year, that was a survivable nuisance. But “under particular conditions and microbiological milieux” it wasn’t. Their models found that population fluctuation, settlement density and wild-animal exploitation — the best available proxies for famine and pathogen load — explain the selection far better than milk consumption does.

Put that in plain terms. Diarrhoea when you are well fed is an unpleasant afternoon. Diarrhoea when you are already starving, or already sick, or living packed in with your animals and your neighbours in a settlement with no sanitation, kills you. The gene didn’t spread because milk was around. It spread because of what happened when milk met a famine.

So the honest version of “why cheese” is not that cheese was the workaround people needed in order to use milk at all. They used milk regardless. What cheese did was make milk keep, travel, and stay safe — which is what let dairying become an economy instead of a seasonal snack. The lactose benefit is real and the excavators said so, but it sits inside a much more interesting story about storage, scale, and hard years.

What I take away from it

I like that this is messier than what I believed. The pouch story is probably true in mechanism and certainly false in detail. The oldest thing we can actually point at is a colander, which makes the simplest cheese in the world — the one I made first, with vinegar, thinking I was cutting corners. And the reason any of it mattered turns out to be less about digestion than about February.

That last part I understand in my bones, because it is the entire argument for the modified refrigerator in my basement. Cheese is how you keep milk. Everything else — the rinds, the moulds, the presses, the two years of patience on a hard wheel — is decoration on top of that one idea.

Up next: the ancient world, where cheese stops being something we infer from residue and starts being something people write down — including the first surviving description of how to actually make it.

Sources

Salque, M., Bogucki, P. I., Pyzel, J., Sobkowiak-Tabaka, I., Grygiel, R., Szmyt, M. & Evershed, R. P. “Earliest evidence for cheese making in the sixth millennium BC in northern Europe.” Nature 493 (2013). https://www.nature.com/articles/nature11698 — paywalled; the University of Bristol’s open summary, which is also where the Bogucki quotation comes from, is at https://www.bristol.ac.uk/news/2012/9008.html. (Cited as 2013 for the print issue; it appeared online in December 2012, which is why you will see both dates.)
Evershed, R. P. et al. “Dairying, diseases and the evolution of lactase persistence in Europe.” Nature 608, 336–345 (11 August 2022). https://www.nature.com/articles/s41586-022-05010-7